Function conv_units() not working inside an ifelse() statement

I want to convert geographic coordinates in a table. I have some measurement in decimal degrees (dec_deg) and some in decimal minutes (deg_dec_min). I want to convert those in decimal minutes to decimal degrees. Using the function conv_unit() within a mutate() and ifelse() statement, I get a Warning message and an incorrect value.

Here's a reproducible example:

library(dplyr)
library(measurements)

data_latlon <- tibble(latitude = c(8.726088, -16.365242, -19.888074, 
                                   '1 40.232', '0 2.308', '2 2.356'),
                      longitude = c(-83.180764, -62.015502, -40.549983, 
                                    '75 54.301', '70 56.693', '72 41.143'), 
                      unit = c('dec_deg', 'dec_deg', 'dec_deg', 
                               'deg_dec_min', 'deg_dec_min','deg_dec_min'))

Case 1: using ifelse()

data_latlon %>% 
  mutate(latitude = ifelse(unit=='deg_dec_min', 
                           conv_unit(latitude, from = 'deg_dec_min', to = 'dec_deg'), 
                           latitude))

# A tibble: 6 × 3
  latitude         longitude  unit       
  <chr>            <chr>      <chr>      
1 8.726088         -83.180764 dec_deg    
2 -16.365242       -62.015502 dec_deg    
3 -19.888074       -40.549983 dec_deg    
4 2.34133333333333 75 54.301  deg_dec_min
5 2.356            70 56.693  deg_dec_min
6 0                72 41.143  deg_dec_min
Warning messages:
1: Problem while computing `latitude = ifelse(...)`.
ℹ longer object length is not a multiple of shorter object length 
2: Problem while computing `latitude = ifelse(...)`.
ℹ data length is not a multiple of split variable 

Case 2: doing it separately

data_latlon %>% 
  filter(unit=='deg_dec_min') %>% 
  mutate(latitude = ifelse(unit=='deg_dec_min', 
                           conv_unit(latitude, from = 'deg_dec_min', to = 'dec_deg'), 
                           latitude))

# A tibble: 3 × 3
  latitude           longitude unit       
  <chr>              <chr>     <chr>      
1 1.67053333333333   75 54.301 deg_dec_min
2 0.0384666666666667 70 56.693 deg_dec_min
3 2.03926666666667   72 41.143 deg_dec_min

I believe the issue is measurments::conv_unit is intended to receive numeric vector. However, some of your values in latitude column is character. This would be an issue if you run the entire column simultaneously.

For example, running conv_unit will return warnings and incorrect results:

conv_unit(data_latlon$latitude, from = 'deg_dec_min', to = 'dec_deg')

[1] "8.99884203333333"  "-19.9047406666667" "-40.232"          
[4] "2.34133333333333"  "2.356"             "0"                
Warning messages:
1: In as.numeric(unlist(strsplit(x_na_free, " "))) * c(3600, 60) :
  longer object length is not a multiple of shorter object length
2: In split.default(as.numeric(unlist(strsplit(x_na_free, " "))) *  :
  data length is not a multiple of split variable

Solution:

The easiest solutions are to vectorize the function or use rowwise():

conv_unit_vec<-Vectorize(conv_unit)


data_latlon %>% 
  mutate(latitude = ifelse(unit=='deg_dec_min', 
                           conv_unit_vec(latitude, from = 'deg_dec_min', to = 'dec_deg'), 
                           latitude))

# A tibble: 6 x 3
  latitude           longitude  unit       
  <chr>              <chr>      <chr>      
1 8.726088           -83.180764 dec_deg    
2 -16.365242         -62.015502 dec_deg    
3 -19.888074         -40.549983 dec_deg    
4 1.67053333333333   75 54.301  deg_dec_min
5 0.0384666666666667 70 56.693  deg_dec_min
6 2.03926666666667   72 41.143  deg_dec_min

or:

data_latlon %>% 
  rowwise()%>%
  mutate(latitude = ifelse(unit=='deg_dec_min', 
                           conv_unit_vec(latitude, from = 'deg_dec_min', to = 'dec_deg'), 
                           latitude)) %>%
  ungroup()

# A tibble: 6 x 3
  latitude           longitude  unit       
  <chr>              <chr>      <chr>      
1 8.726088           -83.180764 dec_deg    
2 -16.365242         -62.015502 dec_deg    
3 -19.888074         -40.549983 dec_deg    
4 1.67053333333333   75 54.301  deg_dec_min
5 0.0384666666666667 70 56.693  deg_dec_min
6 2.03926666666667   72 41.143  deg_dec_min

Few things to note here:

• from ?ifelse : "yes will be evaluated if and only if any element of test is true, and analogously for no"; so both yes and no are fully evaluated here.

• conv_unit() does not really check if values in x arg make sense.
  Snippet from function source:

    if (from == "deg_dec_min") 
      secs = lapply(split(as.numeric(unlist(strsplit(x_na_free, 
        " "))) * c(3600, 60), f = rep(1:length(x_na_free), 
        each = 2)), sum)

Note how it uses unlist(), c(3600, 60) and rep(..., each = 2), it relies on an assumption that each element in input vector x will be split into exactly 2 numbers, no more no less.
If you take your input vector, c("8.726088", "-16.365242", "-19.888074", "1 40.232", "0 2.308", "2 2.356"), split strings by " ", and unlist, you will get 9 instead of 12 numbers. This is the reason for the warning and messed up result.

Besides rowwise() you could also handle this by e.g. split() and map_at() to only process rows where deg_dec_min is set:

library(measurements)
library(dplyr, warn.conflicts = FALSE)
library(purrr)

# split to list of tibbles by "unit" column, 
# apply mutate only on "deg_dec_min" part,
# rbind both parts back together
data_latlon %>% 
  split(~ unit) %>% 
  map_at("deg_dec_min", 
         \(x) x %>% mutate(across(ends_with("itude"),
                                  \(coord_col) conv_unit(coord_col, 
                                                         from = 'deg_dec_min', 
                                                         to = 'dec_deg')))) %>% 
  list_rbind()
#> # A tibble: 6 × 3
#>   latitude           longitude        unit       
#>   <chr>              <chr>            <chr>      
#> 1 8.726088           -83.180764       dec_deg    
#> 2 -16.365242         -62.015502       dec_deg    
#> 3 -19.888074         -40.549983       dec_deg    
#> 4 1.67053333333333   75.9050166666667 deg_dec_min
#> 5 0.0384666666666667 70.9448833333333 deg_dec_min
#> 6 2.03926666666667   72.6857166666667 deg_dec_min

Input:

data_latlon <- tibble(latitude = c(8.726088, -16.365242, -19.888074, 
                                   '1 40.232', '0 2.308', '2 2.356'),
                      longitude = c(-83.180764, -62.015502, -40.549983, 
                                    '75 54.301', '70 56.693', '72 41.143'), 
                      unit = c('dec_deg', 'dec_deg', 'dec_deg', 
                               'deg_dec_min', 'deg_dec_min','deg_dec_min'))

^{Created on 2023-05-10 with reprex v2.0.2}